# Non Linear Iterations and convergence doubt!

## Recommended Posts

Hello,

I would like to understand what the solver is doing wile trying to solve a nonlinear problem.

I understand I have stated in NLPARAM to star iteration at 1.0E-2. I have also activated ExpertNL

In the next picture, what is the solver doing each iteration of each load step? what is changing? what Average U, Err. EPI, Measures EWI mean?

WHat is the goal of the solver to notice it is reaching convergence?

##### Share on other sites

OptiStruct solves nonlinear set of equations with an implicit solution technique

• OptiStruct uses the incremental-iterative procedure based on Newton’s method
• The applied load is broken down into small increments
• For each of the small increments of load an approximate solution is obtained
• Several iterations may be required to obtain a sufficiently accurate approximate solution

OptiStruct checks for convergence on three fronts U,P, and W (Displacement, Load and Work)

• Measures error in the nonlinear solution against a predetermined tolerance level
• The default error tolerances for the convergence criteria are 0.001 for displacement ,load & 1.0E-7 for work/strain
• It is not generally recommended to change the default values. However, varying the tolerances or excluding one or more criteria may affect the result accuracy

Nonlinear expert system

• PARAM, EXPERTNL, AUTO (default)/YES/CNTSTB/NO
• The expert system monitors the convergence of nonlinear processes and tries to improve the convergence for poorly converging cases
• “YES” activates an ‘expert system’ that aids in the convergence. Possible actions like performing additional iterations, under-relaxation, automatic adjustment of the load increment, backing off to the last converged solution and retrying. May lead to a large number of nonlinear iterations.
• “AUTO” activates a ‘light’ version of the expert system which facilitates converging nonlinear process in reasonably close to minimum number of iterations.
• CNTSTB” introduces temporary stabilization on contact interfaces to improve nonlinear convergence
Joel Rodarte likes this

##### Share on other sites

Thank you once again Rahul!

## Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

×   Pasted as rich text.   Paste as plain text instead

Only 75 emoji are allowed.